Cell cultivation
Growth media and inoculum preparation
Picochlorum sp. (BPE23), isolated from a saltwater body of Bonaire was pre-cultivated in shake flasks in an orbital shaker incubator (Multitron, Infors HT) with a 12/12 h day/night cycle and a light intensity of 100 μmolph m−2 s−114. The temperature was 30 °C during night and 40 °C during day. Furthermore, the relative humidity of the air in the incubator was set to 60% and enriched with 2% CO2. Cells were cultured in artificial seawater enriched with nutrients and trace elements. Elements were provided at the following concentrations (in g L−1): NaCl, 24.5; MgCl2.6H2O, 9.80; Na2SO4, 3.20; NaNO3 2.12; K2SO4, 0.85; CaCL2·2H2O, 0.80; KH2PO4, 0.23; Na2EDTA·2H2O, 0.105; Na2EDTA, 0.06; FeSO4·7H2O, 0.0396; MnCl2·2H2O, 1.71 × 10−3; ZnSO4·7H2O, 6.60 × 10−4; Na2Mo4·2H2O, 2.42 × 10−4; Co(NO3)2·6H2O, 7.00 × 10−5; NiSO4·6H2O, 2.63 × 10−5; CuSO4·5H2O, 2.40 × 10−5; K2CrO4, 1.94 × 10−5; Na3VO4, 1.84 × 10−5; H2SeO3, 1.29 × 10−5. HEPES (4.77 g L−1) was added for shake flask cultures as a pH buffer. The medium pH was adjusted to 7.4 after and filter sterilized before use. During photobioreactor cultivation, Antifoam B (J.T.Baker, Avantor, USA) was added at a concentration of 0.5 mL L−1 out of a 1% w/w% stock. In addition, 0.168 g L−1 Sodium bicarbonate (NaHCO3) was added at the time of inoculation to provide sufficient CO2 at the start of the cultivation. The photobioreactor was inoculated at a starting OD density of 0.2.
Photobioreactor operation
Microalgae were cultivated in heat-sterilized flat panel photobioreactors (Labfors 5 Lux, Infors HT, Switzerland) with a working volume of 1.8 L, an optical depth of 20.7 mm and a surface area for irradiation of 0.08 m2. Continuous irradiation (24/24 h) was applied from one side by 260 warm white LED lamps at 813 μmolph m−2 s−1 (PAR). To remove variation in gene expression due to the circadian cycle we grew the microalgae under continuous irradiation while maintaining all other growth conditions stable at the same time. The biomass density was maintained at approximately 2.3 g L−1 by continuous light controlled dilution of the cell culture (turbidostat mode). The turbidostat control was set to maintain an outgoing light level of 10 μmolph m−2 s−1 (PAR). The dilution rate of the photobioreactor was logged continuously by weighing the ingoing medium vessel. Compressed air was supplied at a rate of 980 mL min−1. CO2 was provided on-demand by pH-controlled addition. The pH level in the photobioreactor was set at 7. The photobioreactor temperature was set at 30 °C from the start of cell cultivation. When steady state was reached, the temperature was increased to 42 °C in one step, which took approximately 15 min. Samples were taken daily at 0 h, 1 h, 4 h, 8 h, 24 h after the temperature increase, and once a day for every day onwards to monitor changes in cell physiology.
Biomass analysis
Dry weight
Biomass concentration (g L−1) was measured in duplicate by dry weight determination. Empty Whatman glass microfiber filters (θ 55 mm, pore size 0.7 μm) were dried overnight at 95 °C and placed in a desiccator for 2 h. Filters were then weighed and placed in the mild vacuum filtration setup. Cell culture containing 1–10 mg of microalgae biomass was diluted in 25 mL 0.5 M ammonium formate and filtered. The filter was washed twice with 25 mL 0.5 M ammonium formate to remove residual salts. The wet filter was dried overnight at 95 °C, placed in a desiccator for 2 h, and weighed. Biomass concentration was calculated from the difference in filter weight before and after filtration and drying.
Cell volume and number
Cell size and cell number were measured in duplicate with the Multisizer III (Beckman Coulter Inc., USA, 50 μm aperture). Samples were diluted in two steps before analysis, initially by dilution of 5× in fresh medium, followed by dilution of 100× in Coulter Isoton II. Cell volume was then derived from the cell size by assuming that cells were shaped spherical.
Quantum yield
The cell culture’s quantum yield (Fv/Fm), representing the maximum photosynthetic capacity of photosystem II was determined. Cells were measured after dark adaption at room temperature for 15 min (AquaPen-C 100, PSI; excitation light 455 nm (blue), saturating light pulse: 3000 µmol m−2 s−1).
Biomass harvest and lyophilizing
Biomass samples for compositional analysis were taken at the same moment as when offline measurements were performed. Microalgae cells were pelleted by centrifugation at 4000g for 5 min and washed with 0.5 M ammonium formate. The centrifugation/washing cycle was repeated twice more after which the cell pellet was frozen at − 20 °C. Samples were then lyophilised for 24 h and stored at − 20 °C until further processing.
Pigment analysis
Pigment content was determined through extraction and HPLC analysis 34. 10 mg of lyophilized biomass was disrupted by bead beating (Precellys 24, Bertin Technologies, France) at 5000 rpm for three cycles of 90 s with 60-s breaks on ice between each cycle. The extraction was done through five washing steps with methanol containing 0.1% butylhydroxytoluene. Separation, identification and quantification of pigments were performed using a Shimadzu (U)HPLC system (Nexera X2, Shimadzu, Japan), equipped with a pump, degasser, oven (25 °C), autosampler, and photodiode array (PDA) detector. Separation of pigments was achieved using a YMC Carotenoid C30 column (250 × 4.6 mm 5 μm ID) coupled to a YMC C30 guard column (20 × 4 mm, 5 μm ID)(YMC, Japan) at 25 °C with a flow rate of 1 mL min−1. A sample injection volume of 20µL was used. The mobile phases consisted of Methanol (A), water/methanol (20/80 (v/v%)) containing 0.2% ammonium acetate (B), tert-methyl butyl ether (C) (all solvents were purchased at Sigma Aldrich). The elution protocol started with 0–12 min isocratic A:95% B:5% C:0%, with at 12 min a step to A:80%, B:5%, C:15%, followed by a linear gradient 12–30 min to A:30%, B:5%, C:65%, finally followed by a conditioning phase 30–40 min at the initial concentration. Analytical HPLC standards for chlorophyll a, chlorophyll b, β-carotene, canthaxanthin, violaxanthin, antheraxanthin, zeaxanthin, and lutein had a purity of > 99% (Carotenature, Switzerland).
Fatty acid analysis
Fatty acids within the triacylglycerol (TAG) and polar lipids (PL) fraction were quantified through GC-FID analysis according to35. 10 mg of lyophilised biomass was disrupted by bead beating. The fatty acids were extracted from the disrupted biomass in a mixture of chloroform/methanol (1:1.25, v:v) containing Glyceryl tripentadecanoate (C15:0 TAG) (T4257, Sigma-Aldrich) and 1,2-didecanoyl-sn-glycero-3-phospho-(1′-rac-glycerol) (sodium salt) C10:0 PG (840434, Avanti Polar Lipids Inc) as internal standards for the TAG and the PL fraction, respectively. Separation of TAG and PL was done by use of Sep-Pak Vac silica cartridge (6 cc, 1000 mg; Waters). TAG was eluted from the column with a solution of hexane:diethylether (7:1, v:v) and the PL were eluted with a solution of methanol:acetone:hexane (2:2:1, v:v:v). The extracts were methylated for 3 h at 70 °C in methanol containing 5% H2SO4.
Transcriptome analysis
mRNA extraction and sequencing
mRNA-Seq analysis was done for samples taken at 0 h, 1 h, 4 h, 8 h, 24 h and 120 h after the temperature increase from 30 to 42 °C. Biomass was directly put on ice and centrifuged for 5 min at 4255 g at 2 °C. The cell pellet was then immediately frozen in liquid nitrogen and stored at − 80 °C until further processing. mRNA was extracted from ~ 200 ul frozen cell pellet by automated mRNA extraction (Maxwell® 16 LEV simplyRNA, Promega, USA). Extracted mRNA was tested for integrity (Qsep100, GCbiotech, Netherlands) and quantity (Qubit fluorometer, ThermoFisher, USA). Sequencing libraries were prepared using the NEB Next® Ultra™ mRNA Library Prep Kit. Fragments of the mRNA library in the size range of 250–300 bp were sequenced using the Illumina Novoseq PE150 platform, yielding paired-end reads of 150 nt (Novogene, China). The quality of mRNA-Seq reads was assessed using FastQC v0.11.5 36.
Transcript assembly and annotation
Paired-end reads were mapped to the genome of Picochlorum sp. SENEW3 (assembly ASM87641v1) using HISAT2 v2.2.1 with the -very-sensitive pre-set 33,37. Transcript were assembled and predicted using StringTie v2.1.4, which was guided by the structural annotation of Picochlorum sp. SENEW3. StringTie’s prepDE3 python script was used to extract gene counts and predict genes de novo. Functional annotation was initiated by a BLASTP search of the translated Picochlorum sp. (BPE23) coding sequences against the protein sequences of Arabidopsis thaliana with an E-value threshold of 1E−10. Unannotated genes were filtered and orthology inference was conducted using OrthoFinder v2.5.2 against the protein sequences of the microalgae Auxenochlorella protothecoides, Chlorella variabilis, Chlamydomonas reinhardtii, and Helicosporidium sp. Ortholog gene identifiers were then matched to their gene description. Functional annotation was concluded by matching unannotated genes to their inferred domains, as derived from Pico-PLAZA.
Differential expression analysis and GO, and KEGG enrichment analysis
Pairwise differential expression (DE) analysis was performed using the DESeq2 v1.30.0. R package. Sample-level quality control consisted of pairwise correlation clustering, hierarchical clustering, and Principal Component Analysis (PCA). Fold change values were generated on a log2 scale (LFC). Two designs for data display were implemented; first, where a control condition after 0 h was used for each sample to compare DE between stressed and non-stressed growth. Second, where the previous sampling moment was used as a control condition to compare DE over time. Genes with a false discovery rate (FDR) adjusted p value ≤ 0.05 and an LFC > 1 were considered as significantly differentially expressed.
Arabidopsis thaliana gene identifiers that matched to Picochlorum sp. (BPE23) genes were linked to their corresponding DE-analysis results and used for GO and KEGG pathway enrichment analysis38. Enrichment analyses were conducted by use of clusterProfiler v3.18.1. and org.At.tain.db v3.12.0. packages. GO-terms and KEGG pathways with an FDR-adjusted p value ≤ 0.05 and a positive or negative enrichment score were considered as significantly enriched and visualized with the ggplots2 package.
Network interference analysis
Weighted gene co-expression network analysis (WGCNA) was conducted by applying the WGCNA R package v1.6920. The correlation network was inferred from a correlation matrix of normalized counts. The optimal value of power was determined through scale-free topology analysis. The network was restricted to genes with informative connectivity, referring to a connectivity higher than the median connectivity of the entire network. Modules were then constructed with average linkage hierarchical clustering using distances in the topological overlap construction. Networks were constructed in a hybrid adaptive tree with a deepSplit of 1, a power of 12, a minimum cluster size of 30, a cut height of 0.8, and no PAM-like stage filtering. Subsequently, the gene with the highest eigengene-based connectivity was identified for each module and considered the module’s hub gene. Furthermore, modules were annotated with GO-terms and KEGG pathways to infer functional properties.
